What Actually Happened at Edwards
The YFQ-44A Fury took off from Edwards Air Force Base in California, ingested a target track through Anduril's Lattice mission-autonomy software, and fired an AIM-120 Advanced Medium Range Air-to-Air Missile (AMRAAM) at a "digital target" in secluded airspace over the Mojave Desert. The Air Force described the engagement as the first live munition release from any U.S. Collaborative Combat Aircraft, and the first live-fire from a CCA-type drone globally. The launch was beyond-line-of-sight, meaning the operator who commanded the shot was not in visual contact with the aircraft at the moment of release.
The shot followed a three-step test progression the Air Force has used for decades on crewed fighters. The program first demonstrated inert carriage (the aircraft carrying a missile with no warhead) earlier in 2026, then validated the data link between the operator and the weapon, and only then authorised the live-fire release. Gen. Dale White, Direct Reporting Portfolio Manager for Critical Major Weapon Systems, said in a statement: "The progression from inert carriage earlier in the year to this weapon release demonstrates program maturity, allowing us to validate our digital integration models with actual data." The Air Force emphasised the test "validates our digital integration models with actual data" rather than demonstrating autonomous target selection. Mark Shushnar, Anduril's Vice President of Autonomous Airpower, said the test "demonstrated an end-to-end, beyond-line-of-sight strike against a simulated target" rather than a scripted demonstration.
What the Collaborative Combat Aircraft Program Is
CCA is the Air Force's program of record for uncrewed jet-powered aircraft that fly alongside crewed fighters as force multipliers. Increment 1, the initial operating capability, selected two designs on 17 June 2026, four months ahead of schedule: Anduril's FQ-44A Fury (formerly YFQ-44A) and General Atomics' FQ-42A Dark Merlin (formerly YFQ-42A). Both are intended to operate as semi-autonomous wingmen rather than as remotely piloted drones, with on-board AI decision support responding to tactical inputs from a crewed flight lead.
The two designs are not interchangeable. The Dark Merlin has an internal weapons bay facilitating stealthy carriage; the Fury carries stores externally on two hardpoints, one under each wing. The Air Force framed the dual-vendor award as a "hi-lo" mix: a higher-survivability stealthy drone (Dark Merlin) and a lower-cost attritable drone (Fury). Production plans call for at least 150 aircraft across both designs by 2030, with the long-term fleet target of roughly 1,000 spanning multiple future increments.
Inside the YFQ-44A Fury
The Fury descends from Blue Force Technologies' Fury, an adversary-air trainer acquired by Anduril in 2023 with roughly 90 employees and a composite aerostructures capability. The Air Force assigned the YFQ-44A designation on 3 March 2025, announced by Chief of Staff General David Allvin at the Air and Space Forces Association Warfare Symposium. In the designation, Y denotes prototype, F denotes fighter, and Q denotes uncrewed; the production aircraft drops the Y to become FQ-44A. It is the first time the fighter mission-design series has been applied to an uncrewed aircraft, a deliberate signal that the Air Force regards the CCA as a combat aircraft rather than a support drone.
The production fuel capacity, weight, structural limits, and mission equipment have not been officially released. Anduril has declined to confirm the aircraft's dimensions. The figures that circulate trace to the earlier Blue Force Fury design as reported by The War Zone, and they describe an aircraft that was still evolving toward its combat configuration: roughly 20 feet long, 17 feet wingspan, around 5,000 pounds maximum gross weight, a Williams International FJ44-4M turbofan in the 4,000-pound-thrust class, about Mach 0.95 (transonic) maximum speed, about 50,000 feet service ceiling, load limits up to +9 G and -3 G transient with about +4.5 G sustained, and an internal modular payload capacity of up to roughly 400 pounds. These should be read as legacy indicative numbers rather than confirmed production values.
The Autonomy Stack Problem
The most consequential decision in the CCA program concerns the software. The Air Force buys the airframe and the mission-autonomy software as separate things, connected by a government-controlled interface called the Autonomy Government Reference Architecture, or A-GRA. Two software layers are distinguished: basic vehicle autonomy that flies the aircraft, and mission autonomy that interprets objectives, evaluates the tactical picture, coordinates with other aircraft, and manages payloads within human-set limits. The mission-autonomy layer is machine-learning-based software, and its measured performance is not public.
The autonomy software is competed, not awarded once. The Air Force maintains a six-company pool (Anduril, General Atomics, Lockheed Martin, Northrop Grumman, RTX's Collins Aerospace, and Shield AI) and issued initial production options to three of them: Anduril, Collins Aerospace, and Shield AI. A primary mission-autonomy provider is not scheduled to be chosen until summer 2027, after a further competitive period. The award of the airframe contract to Anduril does not automatically install Anduril's Lattice on every FQ-44A. The Air Force has also tied part of the autonomy vendors' award fee to operator feedback and performance.
Fury has already integrated Lattice and Shield AI's Hivemind; Dark Merlin has integrated Collins Aerospace's Sidekick. Following a six-month performance-based evaluation, the service will narrow it down to one or two solutions during the subsequent six-month period. Then a final selection of a singular autonomy solution will follow in summer 2027. This is the part of the program that determines whether CCA becomes a true force multiplier or an expensive target drone: a good airframe with bad software is just a missile magnet.
The Competitive Landscape: Fury vs Dark Merlin vs The Dropouts
The Air Force had earlier declined to down-select for flight testing from Boeing, Lockheed Martin, and Northrop Grumman, all of whom had pitched loyal wingman designs. That said, Northrop's Talon / Project Lotus demonstrator has since received an official prototype designation as YFQ-48A Talon Blue and should flight test this year. The dropouts are not out of the program; they remain competitors for future increments and for the autonomy stack.
| Attribute | FQ-44A Fury (Anduril) | FQ-42 Dark Merlin (General Atomics) |
|---|---|---|
| Design heritage | Blue Force adversary-air Fury, acquired 2023 | Derived from XQ-67A demonstrator |
| First flight | 31 October 2025 | August 2025 |
| Live-fire milestone | Live AIM-120 fired July 2026 | Planned by end of 2026 |
| Flight-test loss | None publicly reported | Total-loss crash April 2026; returned to flight May 2026 |
| Weapons carriage | External hardpoints under each wing | Internal weapons bay (stealth) |
| Engine | Williams International FJ44-4M turbofan (~4,000 lbf) | Undisclosed |
| Combat radius requirement | >700 nautical miles (Increment 1) | >700 nautical miles (Increment 1) |
| Autonomy stack | Lattice + Hivemind | Sidekick |
| Open autonomy architecture | Yes (A-GRA compatible) | Yes (A-GRA compatible) |
| Unit cost (planning) | ~$30M (one-third F-35A) | ~$30M (one-third F-35A) |
| Production site | Arsenal-1, Pickaway County, Ohio | Undisclosed |
The most consequential operational distinction between the two: the Dark Merlin's internal weapons bay enables stealth operations, which should make it more survivable in contested airspace. If the Fury is less expensive to acquire and operate (which is not confirmed), the service may desire a mix: higher-cost stealthy CCAs for the first salvo against defended targets, lower-cost CCAs for the mass follow-on waves. This is the same attritable-aircraft logic the Air Force used to justify the original Reaper versus Predator mix.
What This Actually Means
The YFQ-44A live-fire closes one chapter in weapons integration testing while leaving others open: the YFQ-42A's planned live-fire later this year, further operational validation, and the ramp-up of weapons and other testing as the Air Force pursues initial fielding by decade's end. For now, the Mojave Desert shot stands as the first U.S. CCA AIM-120 launch and a tangible data point in the program's push from captive-carry and simulation to operational capability.
What the test proved: that an autonomous fighter jet can carry a live air-to-air missile, receive a target track through the Lattice software, accept a human command to engage, and execute the launch within pilot-defined parameters. What the test did not prove: that the system can autonomously select and engage targets in denied environments, that the A-GRA interface holds up under jamming or datalink loss, that the unit cost will clear the planning target, that the autonomy stack will pass the summer 2027 down-select, or that the Air Force can sustain production at the required rate. The July shot validated a controlled sequence against a digital target. Combat capability is a different milestone.
The harder claims are the ones the public record cannot yet support: an affordable unit price, a survivable combat radius, a defined sensor fit, the autonomy stack down-select, or superiority over the FQ-42. Each is undisclosed, and the July live-fire validated a single controlled sequence against a digital target rather than the combat capability the Air Force is ultimately buying. The FQ-44A has crossed from technology demonstration into an early operational weapon-system program, but the harder milestones remain ahead.
Sources
- Anduril — Fury product page — official description of Fury platform, Lattice integration, Arsenal-1 production site.
- Defence Blog — U.S. Air Force conducts live-fire test for its unmanned fighter jet — primary source for the July 2026 AIM-120 launch from the YFQ-44A.
- The War Zone — YFQ-44A Fury Fighter Drone Has Fired Its First AIM-120 AMRAAM Missile — primary source for the technical context of the test and the three-step progression.
- Defense News — U.S. Air Force Fires First Live AIM-120 Missile from YFQ-44A Collaborative Combat Aircraft — primary source for Mark Shushnar and Gen. Dale White statements.
- Drone Warfare — FQ-44A Fury: Anduril's CCA Drone — primary source for design heritage, autonomy stack competition, and the Y-prefix designation history.
- Inside Unmanned Systems — Air Force Goes Split Buy — primary source for the dual-vendor June 2026 award and the hi-lo mix rationale.
- OINTS Insights — Anduril's YFQ-44A Drone Fires First Live AIM-120 Missile — primary source for the beyond-line-of-sight engagement description and the test methodology.
- TechRadar — USAF's Autonomous AI fighter drone just fired a real AMRAAM — primary source for the FY2027 budget breakdown ($996.5M procurement, $1.37B R&D).
- GovConFeed — Air Force Awards Anduril and General Atomics Production Contracts June 2026 — primary source for the procurement contract details and the 150-aircraft-by-2030 target.
Frequently Asked Questions
Is the YFQ-44A Fury autonomous?
The Fury has full autonomous flight and manoeuvring capability, but the Air Force has specifically stated that the aircraft is not authorised to make autonomous decisions to use lethal force. During the July 2026 test, a human operator received the target track, evaluated the engagement, and issued the command to launch. The autonomy stack (Lattice or Hivemind) handles navigation, sensor fusion, and tactical coordination within pilot-defined parameters, but the lethal-force decision remains with a human operator.
What is the AIM-120 AMRAAM?
The AIM-120 Advanced Medium Range Air-to-Air Missile is the U.S. military's primary beyond-visual-range air-to-air missile. It uses active radar homing and has a range of roughly 20-100 miles depending on the variant. It is the standard weapon for F-15, F-16, F/A-18, and F-22 aircraft, and now, with the CCA test, for an autonomous drone as well.
Why does the Air Force want autonomous wingmen?
The operational logic is force multiplication: one crewed fighter (F-35, F-22, F-15EX) leads two or three CCAs, expanding the sensor reach and weapons capacity without putting more pilots at risk. The Air Force estimates that CCAs cost roughly one-third the price of an F-35A (~$30M vs. ~$83M), enabling procurement at scale. The attritable-aircraft logic means CCAs can be sent into contested airspace where sending crewed aircraft would carry unacceptable risk.
What is the A-GRA interface?
The Autonomy Government Reference Architecture is a government-controlled interface that separates the airframe from the mission-autonomy software. It allows different software providers (Anduril, Collins Aerospace, Shield AI) to operate on the same airframe. This is the mechanism by which the Air Force can compete the autonomy stack separately from the airframe contract, and is what made the February 2026 dual-stack flight test possible.
When will CCA be operational?
The Air Force has set a target of at least 150 combat-capable CCAs by the end of the decade, with the long-term fleet target of roughly 1,000 aircraft spanning multiple future increments. Initial fielding is expected before 2030, but the Air Force has not specified an exact date for first operational deployment. The program is currently in early production, with both vendors beginning manufacturing capacity ramp-up.
Could the CCA launch a missile without human authorisation?
Not under current policy. The Air Force has been explicit that the human operator remains the lethal-force decision authority. The autonomy stack handles target detection, tracking, and tactical coordination, but the actual launch command requires a human authorisation. The autonomy stack could in principle be configured to engage automatically within human-set parameters, but this is policy-dependent and not currently authorised for the CCA program.
How does CCA compare to drones like MQ-9 Reaper?
The Reaper is a remotely piloted aircraft with a ground-based pilot who controls most functions. CCA is a semi-autonomous aircraft that flies itself with on-board AI decision support and responds to tactical inputs from a crewed flight lead. The CCA also operates in a different regime: jet-powered, transonic, designed for air-to-air combat in contested airspace, rather than turboprop-powered, subsonic, designed for long-endurance ISR and strike missions in permissive airspace.
Loading comments…